Unsmoothing smoothed return series for risk management and asset allocation

Abstract

Smoothed returns that can occur as a result of an appraisal-based valuation process distort asset characteristics and, in particular, volatility and correlations. Using a simple but often-used model for smoothed return series, this article analyzes what unsmoothing techniques can achieve with respect to uncovering the true time-series properties. The properties of interest are volatility, autocorrelation, self-correlation, and cross-asset cor- relation. It is shown that, specifically for heavily smoothed return series, volatility estimates might overstate the true volatility, and cross-asset correlations cannot fully be recovered.